# Requirements for PII extraction script
# Install with: pip install -r requirements.txt

# Microsoft Presidio - PII detection using NER and pattern matching
presidio-analyzer>=2.2.0,<3.0.0

# spaCy - NLP library providing NER models for Presidio
spacy>=3.7.0,<4.0.0

# HuggingFace datasets - Streaming access to training datasets
datasets>=2.14.0,<4.0.0

# HuggingFace Hub - Authentication and dataset access
huggingface_hub>=1.0

# Note: After installing, download the spaCy English model:
#   python -m spacy download en_core_web_lg
